A method of forming a mosfet device is provided. The method includes providing a substrate. The method includes forming on the substrate a relaxed sige layer having a ge content between 0.51 and 0.80. Furthermore, the method includes depositing on the relaxed sige layer a ε-Si layer.

Patent
   7005668
Priority
Jun 25 2002
Filed
Jun 25 2003
Issued
Feb 28 2006
Expiry
Jun 25 2023
Assg.orig
Entity
Small
0
7
all paid
1. A mosfet device comprising:
a substrate;
a relaxed sige layer that is formed on said substrate, said relaxed sige layer having a ge content between 0.51 and 0.80 and a selective portion having a ge content between 0.7 and 0.75; and
a ε-Si layer that is deposited on said relaxed sige layer.
26. A mosfet device comprising:
a substrate;
a relaxed sige layer that is formed on said substrate, said relaxed sige layer having a ge content between 0.51 and 0.80 and a selective portion having a ge content between 0.7 and 0.75; and
a ε-Si layer that is deposited on said relaxed sige layer so that hole mobility enhancement increases with effective vertical field.
14. A mosfet device comprising:
a substrate;
a relaxed sige layer that is formed on said substrate, said relaxed sige layer having a ge content between 0.51 and 0.80; and
a digital alloy structure that is formed on said relaxed sige layer comprising alternating layers of ε-Si and sige having a ge content between 0.51 and 1, wherein said mobility enhancement of said device is constant.
2. The mosfet device of claim 1, wherein said ε-Si layer is sized approximately at 45 Å.
3. The mosfet device of claim 1, wherein said relaxed sige layer is planarized.
4. The mosfet device of claim 1 further comprising a hole mobility enhancement that increases with effective vertical field.
5. The mosfet device of claim 4, wherein said hole mobility enhancement saturates approximately around 2.5.
6. The mosfet device of claim 1, wherein said ε-Si layer shifts the hole wave function away from the surface of said ε-Si layer.
7. The mosfet device of claim 1, wherein said substrate comprises a crystalline Si substrate.
8. The mosfet device of claim 1, wherein said substrate comprises a crystalline Si substrate and a relaxed sige graded layer.
9. The mosfet device of claim 1, wherein said substrate comprises a crystalline substrate and an insulating layer.
10. The mosfet device of claim 9, wherein said insulator layer comprises an oxide.
11. The mosfet device of claim 1 further comprising a PMOS device.
12. The mosfet device of claim 11 further comprising a NMOS device.
13. The mosfet device of claim 12, wherein said PMOS and NMOS devices form a CMOS device.
15. The mosfet device of claim 14, wherein said alternating layers of sige and ε-Si is sized approximately at 45 Å.
16. The mosfet device of claim 14, wherein said relaxed sige layer is planarized.
17. The mosfet device of claim 14, wherein said ε-Si layer shifts the hole wave function away from the surface of said ε-Si layer.
18. The mosfet device of claim 14, wherein said substrate comprises a crystalline Si substrate.
19. The mosfet device of claim 14, wherein said substrate comprises a crystalline Si substrate and a relaxed sige graded layer.
20. The mosfet device of claim 14, wherein said substrate comprises a crystalline substrate and an insulating layer.
21. The mosfet device of claim 20, wherein said insulator layer comprises an oxide.
22. The mosfet device of claim 14 further comprising a PMOS device.
23. The mosfet device claim 22 further comprising a NMOS device.
24. The mosfet device claim 23, wherein said PMOS and NMOS devices form a CMOS device.
25. The mosfet device of claim 14, wherein said relaxed sige layer comprises a selective portion having a ge content between 0.7 and 0.75.
27. The mosfet device of claim 26, wherein said ε-Si layer is sized approximately at 45 Å.
28. The mosfet device of claim 26, wherein said relaxed sige layer is planarized.
29. The mosfet device of claim 26, wherein said hole mobility enhancement saturates approximately around 2.5.
30. The mosfet device of claim 26, wherein said ε-Si layer shifts the hole wave function away from the surface of said ε-Si layer.
31. The mosfet device of claim 26, wherein said substrate comprises a crystalline Si substrate.
32. The mosfet device of claim 26, wherein said substrate comprises a crystalline Si substrate and a relaxed sige graded layer.
33. The mosfet device of claim 26, wherein said substrate comprises a crystalline substrate and an insulating layer.
34. The mosfet device of claim 33, wherein said insulator layer comprises an oxide.
35. The mosfet device of claim 26 further comprising a PMOS device.
36. The mosfet device of claim 35 further comprising a NMOS device.
37. The mosfet device claim 36, wherein said PMOS and NMOS devices form a CMOS device.

This application claims priority from provisional application Ser. No. 60/391,452 filed Jun. 25, 2002, which is incorporated herein by reference in its entirety.

The invention relates to the field of p-type MOSFETS, and in particular to improving hole mobility in strained silicon p-type MOSFETS.

Strained silicon grown on relaxed Si1-xGex virtual substrates has been used to fabricate both n- and p-type MOSFETs, which exhibit enhanced carrier mobility compared to bulk silicon. Biaxial tensile strain breaks the six-fold degeneracy of silicon's conduction band, resulting in reduced intervalley scattering in nMOS devices. Furthermore, for in-plane transport, electrons have only the low transverse effective mass (mt=0.19m0). For ε-Si grown on Si0.8Ge0.2, the conduction band splitting is large enough to completely suppress intervalley scattering, and no further improvement in electron mobility is gained by increasing the strain in the silicon layer.

Biaxial tensile strain also splits the light-hole/heavy-hole degeneracy in the valence band. Unlike the conduction band, strain also changes the shape of the light-hole valley, resulting in lower in-plane and out-of-plane effective masses. Since the rate of subband splitting in the valence band is known to be lower than for the conduction band, theory predicts that intervalley scattering for holes will not be suppressed until the strain reaches 1.6%, corresponding to growth on a Si0.6Ge0.4 buffer. Recent experimental work shows that hole mobility enhancement saturates for ε-Si on Si0.6Ge0.4, with no further improvement as the virtual substrate Ge content was increased to 50%.

Unlike ε-Si nMOS, mobility enhancements in ε-Si p-type MOSFETs demonstrate a functional dependence on vertical effective field. While an 80% electron mobility enhancement has been observed in ε-Si for vertical fields ranging from 0.1 to 1MV/cm, hole mobility enhancements tend to evolve as the effective field changes, as shown in FIG. 1. For low fields, the hole mobility enhancement typically increases as the vertical field increases. At an intermediate field value, the enhancement peaks and then degrades with further increases in field.

According to one aspect of the invention, there is provided a method of forming a MOSFET device. The method includes providing a substrate. The method includes forming on the substrate a relaxed SiGe layer having a Ge content between 0.51 and 0.80. Furthermore, the method includes depositing on the relaxed SiGe layer a ε-Si layer.

According to another aspect of the invention, there is provided a method of forming a MOSFET device. The method includes providing a substrate. Also, the method includes forming on the substrate a relaxed SiGe layer having a Ge content between 0.51 and 0.80. Furthermore, the method includes forming on the relax SiGe layer a digital alloy structure that comprises alternating layers of ε-Si and SiGe having a Ge content between 0.51 and 1 so that the mobility enhancement of the device is constant.

According to another aspect of the invention, there is provided a method if forming a MOSFET device. The method includes providing a substrate. The method includes forming on the substrate a relaxed SiGe layer having a Ge content between 0.51 and 0.80. Furthermore, the method includes depositing on the second SiGe buffer a ε-Si layer so that hole mobility enhancement increases with effective field.

According to another aspect of the invention, there is a provided a MOSFET device. The MOSFET device includes a substrate. A relaxed SiGe layer is formed on the substrate having a Ge content between 0.51 and 0.80. A ε-Si layer is deposited on the relaxed SiGe layer.

According to another aspect of the invention, there is a provided a MOSFET device. The MOSFET device includes a substrate. A relaxed SiGe layer is formed on the substrate having a Ge content between 0.51 and 0.80. A digital alloy structure is formed on the relaxed SiGe layer comprising alternating layers of ε-Si and SiGe having a Ge content between 0.51 and 1. The mobility enhancement of the device is constant.

According to another aspect of the invention, there is a provided a MOSFET device. The MOSFET device includes a substrate. A relaxed SiGe layer is formed on the substrate having a Ge content between 0.51 and 0.80. A ε-Si layer is deposited on the relaxed SiGe layer so that hole mobility enhancement increases with effective vertical field.

FIG. 1 is a graph demonstrating mobility enhancement as a function of vertical effective field in ε-Si p-type MOSFETs;

FIG. 2A is a XVTEM diagram demonstrating a ε-Si layer grown on a Si0.3Ge0.7 buffer; FIG. 2B is a graph demonstrating mobility enhancement of ε-Si layer on Si0.3Ge0.7 pMOSFET compared to other high-mobility ε-Si pMOS structures; and

FIG. 3A is a XVTEM diagram demonstrating a ε-Si/Si0.3Ge0.7 digital alloy structure grown on a Si0.3Ge0.7 buffer;

FIG. 3B is a graph demonstrating mobility enhancement of ε-Si/Si0.3Ge0.7 digital alloy pMOSFET compared to a convention ε-Si structure.

While the benefit gained from suppression of intervalley scattering appears to be saturated in ε-Si pMOS, further boosts in hole mobility enhancement should be possible by continuing to increase the Ge content in the relaxed SiGe buffer. This is due to the effective mass of holes in the vertical direction is very low, meaning that some part of the hole's wave function is likely to be present below the ε-Si surface channel, even at high vertical fields. Since the band structure of Si1-xGex rapidly becomes Ge-like at x=0.7 to 0.75, the hole effective mass in a relaxed Si1-xGex alloy likewise starts to resemble the low hole effective mass measured in pure Ge. The invention includes techniques to increase the hole mobility enhancement in ε-Si pMOS as well as methods for controlling the enhancement as a function of vertical effective field.

Relaxed graded SiGe buffers are grown on a crystalline Si substrate to a Ge content of 60% via ultrahigh vacuum chemical vapor deposition (UHVCVD). In other embodiments, the substrate can be a crystalline Si substrate and a relaxed SiGe graded layer, a crystalline substrate and an insulating layer, or the like. The wafers are then removed from the system and subjected to chemo-mechanical polishing (CMP) to remove crosshatch surface roughness and to reduce the density of dislocation pileups. At this point, the relaxed SiGe layer may be transferred to an alternative substrate, such as an SOI wafer, or left as-is. After recleaning, the wafers are reinserted into the UHVCVD for further grading and device layer deposition. A structure 2 is grown comprising of a 45 Å ε-Si layer 4 on a Si0.3Ge0.7 relaxed graded buffer 6, as shown in FIG. 2A, and fabricated into long-channel MOSFETs, which utilize a deposited low temperature oxide as the gate dielectric and a single mask level for device patterning. The buffer 6 can have Ge concentrations between 0.51 and 0.8 for effective operation of the invention.

The inversion layer effective mobility is extracted from the linear regime I–V characteristics. The hole mobility enhancement is a function of vertical effective field for this structure, as shown in FIG. 2B, along with other high mobility ε-Si pMOSFETs. As can be seen, the hole mobility enhancement in the ε-Si on Si0.3Ge0.7 pMOSFET increases with effective vertical field over a wide field range and saturates at a value of around 2.5. This novel behavior is useful, because it allows for improved pMOSFET performance in deeply scaled devices with large vertical fields. When the inversion layer first forms at low vertical effective fields in this device, the hole wave function is weighted towards the Ge-rich relaxed buffer because of the large valence band offset between strained Si and SiGe (type II alignment).

Mobility enhancement in this regime probably results from the low effective mass of holes in the Si0.3Ge0.7 compared to bulk Si. However, because the Si0.3Ge0.7 is relaxed, its valence band is degenerate and intervalley scattering is present as a mobility limiting mechanism. As the gate overdrive is increased, the centroid of the wave function is pulled closer and closer to the surface. This shifting of the hole wave function towards the surface ε-Si layer adds the benefit of valence band splitting, resulting in mobility enhancements exceeding those previously seen in ε-Si p-type MOSFETs. Even as the wave function approaches the surface, a significant portion of the wave function's tail should always be present in the relaxed Si0.3Ge0.7, because the Si cap is so thin. In this particular demonstration, the Si cap was 45 Å thick, thinner layers provide similar benefits. If the Si cap is grown thick enough to contain all or most of the hole wave function, the hole will lose contact with the Si0.3Ge0.7 at high vertical field and some enhancement will be lost. N-MOSFETs with enhanced electron mobility can also be fabricated on the same wafers, making these materials suitable for enhanced-performance CMOS applications.

A second structure 8 is grown on a relaxed Si0.3Ge0.7 buffer, except that instead of simply capping the structure with ε-Si, the digital alloy 10 consisting of 7 periods of alternating ε-Si 12 and Si0.3Ge0.7 14 layers is grown, as shown in FIG. 3A. Each layer 12, 14 is approximately 8 Å thick, and the entire structure 8 is capped with 20 Å of ε-Si in order to allow the use of a SiO2 gate. When the inversion layer forms in this structure 8, the hole should have enhanced transport from the splitting of the valence band degeneracy as long as the hole wave function is within 120 Å of the surface. The SiGe layers 14 can have Ge concentrations between 0.51 and 0.8.

As can be seen in FIG. 3B, the mobility enhancement in this device demonstrates no dependence on vertical effective field. This represents a substantial improvement over prior art ε-Si p-MOSFETs, where a constant mobility enhancement had previously been thought to be unattainable. Thus, unlike the conventional ε-Si p-type MOSFET, the digital alloy device does not need to be biased to high vertical field in order for the hole wave function to combine reduced intervalley scattering with the Ge-like effective mass of Si0.3Ge0.7. In general, periodic, repeating layer structures (such as the digital alloy described here) are useful for allowing the hole mobility enhancement to be fixed at a constant value with respect to effective field. n-MOSFETs with electron mobility comparable to Cz-Si can also be fabricated from the same digital-alloy material.

It is important to note that despite the fact that the individual layers 12, 14 are extremely thin, the hole is not experiencing the valence band structure of a random alloy. If the valence band structure is that of the average composition of the layers, then the digital alloy could be replaced by a tensile Si0.65Ge0.35 layer on a Si0.3Ge0.7 buffer. However, according to recent alloy scattering studies, such a structure would actually exhibit hole mobility below that of bulk Si.

Even though valence band splitting in ε-Si saturates for buffer compositions greater than 40% Ge, further mobility enhancements in ε-Si p-type MOSFETs are possible through the use of a high Ge content relaxed buffer. The large enhancements seen at high vertical fields result from a hybrid of the valence band splitting present in the ε-Si cap and the Ge-like effective mass in the Si0.3Ge0.7 buffer. Since the ε-Si layer is only 45 Å thick, the hole wave function can always sample the relaxed Si0.3Ge0.7 buffer, even as the hole wave function is pulled towards the surface by the vertical field. The digital alloy 10 is a structure designed and grown consisting of alternating layers of Si0.3Ge0.7 and ε-Si upon a Si0.3Ge0.7 buffer. The alternating layer structure 10 allows the hole wave function to sample both the low effective mass in Si0.3Ge0.7 and the valence band splitting in the ε-Si at both low and high vertical field. The use of different compositions in the alternating layers can also lead to greatly enhanced hole mobility. Even though the layers 12, 14 comprising the digital alloy are on the order of several atomic layers thick, the hole is able to combine the unique benefits intrinsic to each “digit” of the alloy.

Although the present invention has been shown and described with respect to several preferred embodiments thereof, various changes, omissions and additions to the form and detail thereof, may be made therein, without departing from the spirit and scope of the invention.

Fitzgerald, Eugene A., Lee, Minjoo L.

Patent Priority Assignee Title
Patent Priority Assignee Title
5426316, Dec 21 1992 International Business Machines Corporation Triple heterojunction bipolar transistor
5906951, Apr 30 1997 GLOBALFOUNDRIES Inc Strained Si/SiGe layers on insulator
6350993, Mar 12 1999 GLOBALFOUNDRIES Inc High speed composite p-channel Si/SiGe heterostructure for field effect devices
6399502, Mar 31 1999 Fahrenheit Thermoscope LLC; Fahrenheit Thermoscope, LLC Process for fabricating a planar heterostructure
6562703, Mar 13 2002 Sharp Laboratories of America, Inc.; Sharp Laboratories of America, Inc Molecular hydrogen implantation method for forming a relaxed silicon germanium layer with high germanium content
6805962, Jan 23 2002 GLOBALFOUNDRIES U S INC Method of creating high-quality relaxed SiGe-on-insulator for strained Si CMOS applications
20020052084,
///
Executed onAssignorAssigneeConveyanceFrameReelDoc
Jun 25 2003Massachusetts Institute of Technology(assignment on the face of the patent)
Nov 05 2003LEE, MINJOO L Massachusetts Institute of TechnologyASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS 0146800131 pdf
Nov 05 2003FITZGERALD, EUGENE A Massachusetts Institute of TechnologyASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS 0146800131 pdf
Date Maintenance Fee Events
Aug 28 2009M2551: Payment of Maintenance Fee, 4th Yr, Small Entity.
Feb 02 2010ASPN: Payor Number Assigned.
Feb 02 2010RMPN: Payer Number De-assigned.
Aug 28 2013M2552: Payment of Maintenance Fee, 8th Yr, Small Entity.
Aug 28 2017M2553: Payment of Maintenance Fee, 12th Yr, Small Entity.


Date Maintenance Schedule
Feb 28 20094 years fee payment window open
Aug 28 20096 months grace period start (w surcharge)
Feb 28 2010patent expiry (for year 4)
Feb 28 20122 years to revive unintentionally abandoned end. (for year 4)
Feb 28 20138 years fee payment window open
Aug 28 20136 months grace period start (w surcharge)
Feb 28 2014patent expiry (for year 8)
Feb 28 20162 years to revive unintentionally abandoned end. (for year 8)
Feb 28 201712 years fee payment window open
Aug 28 20176 months grace period start (w surcharge)
Feb 28 2018patent expiry (for year 12)
Feb 28 20202 years to revive unintentionally abandoned end. (for year 12)